Publication: Molecular subtyping in colon cancer through consensus clustering of multi-omics data
Abstract
Kolon kanseri, dünya çapında kansere bağlı ölümlerin en yaygın nedenlerinden biridir ve en yaygın histolojik alt tiplerinden biri kolon adenokarsinomudur. Kolon kanseri hastaları, hastalığın çok çeşitli heterojen yapısından dolayı mevcut tedavi stratejilerinden yeterince yararlanamamaktadır. Kanserin moleküler alt tiplendirilmesi, kanserin heterojenliği hakkında derinlemesine bilgi sağlayabilir ve en uygun hedefe yönelik tedavi stratejilerinin seçimine katkıda bulunabilir. Bu çalışmanın temel amacı, çoklu omik verilere ve bütünleştirici kümeleme algoritmalarına dayalı konsensüs kümeleri oluşturarak kolon kanserinin kapsamlı moleküler karakterizasyon analizini gerçekleştirmektir. Ayrıca hastalar sistematik olarak sınıflandırılarak spesifik moleküler özellikleri belirlenerek hassas onkoloji uygulamalarının önü açıldı. Bu tez çalışmasında kolon adenokarsinomu, R paketi MOVICS kullanılarak multi-omik verilere ve 10 bütünleştirici kümeleme algoritmasına dayalı konsensüs kümeleri oluşturularak sistematik olarak sınıflandırılmış ve tanımlanan alt tipler çeşitli açılardan incelenerek kişiye özel tedavi seçenekleri değerlendirilmiştir. Sağlam bir kümeleme çıktısı elde etmek için doğrulama kohortunda eş zamanlı olarak alt tipleme analizleri de yapıldı ve her iki analizin sonuçları tutarlılık ve tekrarlanabilirlik açısından değerlendirildi. Sonuç olarak COAD'ın yüksek hassasiyetle tespit edilen 2, 3 ve 4 alt kümesi ayrı ayrı değerlendirildi. Her üç değerlendirme sonucunda alt tiplerin mRNA, lncRNA, miRNA, metilasyon ve mutasyon açısından farklılık gösterdiği belirlendi. Özellikle APC mutasyonu her üç değerlendirmede de en sık mutasyona uğrayan gendi. Evreye bağlı farklılıklar sadece 4 alt tipleme çalışmasında görüldü. Her üç değerlendirme sonucunda 5-fluorouracil, cisplatin, paclitaxel, SN-38, rapamycin ve imatinib ilaç tedavilerine istatistiksel olarak anlamlı yanıt olduğu belirlendi. Ayrıca validasyon analizleri ile analizlerin doğruluğu teyit edilmiştir.
Colon cancer is one of the most common causes of cancer-related deaths worldwide, and one of the most common histological subtypes is adenocarcinoma of the colon. Colon cancer patients cannot benefit sufficiently from current treatment strategies due to the great diversity and heterogeneity of the disease. Molecular subtyping of cancer can provide profound information about the heterogeneity of the cancer and contribute to the selection of the most appropriate targeted treatment strategies. The main objective of this thesis study was to perform a comprehensive molecular characterization analysis of colon cancer by creating consensus clusters based on multi omics data and integrative clustering. In addition, patients were systematically classified, their specific molecular characteristics were determined and the way for precision oncology applications has been paved. In this thesis study, adenocarcinoma of the colon was systematically classified by creating consensus clusters based on multi-omics data and 10 integrative clustering algorithms using the R package MOVICS, and personalized treatment options were evaluated by examining the defined subtypes with respect to various parameters. In order to obtain a robust clustering output, subtyping analyses were also performed simultaneously in the validation cohort and the results of both analyses were evaluated for consistency and reproducibility. In conclusion, 2, 3 and 4 subsets of COAD, which were detected with high sensitivity, were evaluated separately. As a result of all three evaluations, it was determined that the subtypes differed in terms of mRNA, lncRNA, miRNA, methylation and mutation. In particular, APC mutation was the most frequently mutated gene in all three evaluations. Differences depending on stage were seen only in 4 subtyping studies. As a result of all three evaluations, it was determined that there was a statistically significant response to 5-fluorouracil, cisplatin, paclitaxel, SN-38, rapamycin and imatinib drug treatments. In addition, the accuracy of the analyzes was confirmed by validation analyses.
Colon cancer is one of the most common causes of cancer-related deaths worldwide, and one of the most common histological subtypes is adenocarcinoma of the colon. Colon cancer patients cannot benefit sufficiently from current treatment strategies due to the great diversity and heterogeneity of the disease. Molecular subtyping of cancer can provide profound information about the heterogeneity of the cancer and contribute to the selection of the most appropriate targeted treatment strategies. The main objective of this thesis study was to perform a comprehensive molecular characterization analysis of colon cancer by creating consensus clusters based on multi omics data and integrative clustering. In addition, patients were systematically classified, their specific molecular characteristics were determined and the way for precision oncology applications has been paved. In this thesis study, adenocarcinoma of the colon was systematically classified by creating consensus clusters based on multi-omics data and 10 integrative clustering algorithms using the R package MOVICS, and personalized treatment options were evaluated by examining the defined subtypes with respect to various parameters. In order to obtain a robust clustering output, subtyping analyses were also performed simultaneously in the validation cohort and the results of both analyses were evaluated for consistency and reproducibility. In conclusion, 2, 3 and 4 subsets of COAD, which were detected with high sensitivity, were evaluated separately. As a result of all three evaluations, it was determined that the subtypes differed in terms of mRNA, lncRNA, miRNA, methylation and mutation. In particular, APC mutation was the most frequently mutated gene in all three evaluations. Differences depending on stage were seen only in 4 subtyping studies. As a result of all three evaluations, it was determined that there was a statistically significant response to 5-fluorouracil, cisplatin, paclitaxel, SN-38, rapamycin and imatinib drug treatments. In addition, the accuracy of the analyzes was confirmed by validation analyses.
Description
Keywords
Bioengineering, Bioinformatics, Biyoinformatik, Biyomühendislik, Colon cancer, Colonic neoplasms, Consensus Ensemble, Çoklu-omik Veri Analizi, Hassas Onkoloji, Kişiselleştirilmiş Tedavi, Kolon kanseri, Kolon tümörleri, Konsensus Kümeleme, Multi-omics Data Analysis, Personalized Treatment, Precision Oncology, Sistem Biyolojisi, System Biology